Case Study 50 Allergic Asthma Frank Morgan

Introduction

Respiratory syncytial virus (RSV) infection is the most common cause of severe bronchiolitis in infants.1 It is now established that early severe viral lower respiratory tract infections (LRTI) are frequently followed by recurrent wheeze, asthma and allergy during childhood.23

A number of previous studies have investigated the association between RSV infection in infancy and later development of asthma, although differences in timing and severity of RSV infection exist between studies. The Tucson Children's Respiratory Study (a birth cohort study) reported an increased prevalence of respiratory symptoms until age 11 and residual impaired lung function at age 13 in those experiencing a relatively mild RSV LRTI (not requiring hospitalisation) during the first 3 years of life.4 Impaired lung function at age 10 has also been reported following hospitalisation in the first year with proven LTRI.5 Recently, data from a cohort previously hospitalised in the first 2 years of life and followed up at 18–20 years suggested that this impairment may persist into early adulthood.6 Histopathology studies of fatal RSV bronchiolitis demonstrate extensive damage to the airway epithelium and marked small airway obstruction,7 yet the presence of persistent small airway damage in later life is unclear. In adult asthma, small airway dysfunction is closely associated with bronchial hyper-responsiveness,89 a fundamental component of asthma correlated to disease severity.10

Our cohort represents infants aged <1 year with severe primary RSV bronchiolitis (43/47 ≤6 months of age at hospitalisation). We have previously demonstrated increased prevalence of asthma or recurrent wheeze (RW) and allergic sensitisation compared with a matched control cohort at ages 3, 7 and 13 years.11–13 In this paper we present the 18-year follow-up data together with a longitudinal analysis of wheeze and allergy patterns. The primary aim was to see if the higher prevalence of asthma or RW, clinical allergy and allergic sensitisation persist at age 18. The secondary aims were to describe the wheeze patterns in both cohorts and to investigate how large and small airway function relate to severe RSV bronchiolitis, current asthma, airway hyper-responsiveness (AHR) and markers of allergic inflammation.

Methods

The study design is summarised in table 1. Forty-seven children aged <1 year hospitalised with RSV LRTI14 between December 1989 and April 1990 constituted the index group. An age- and gender-matched control group (n=93) was recruited from children attending the same child healthcare centres as the index cases. Both cohorts were followed up prospectively at ages 1, 3, 7, 13 and 18 years. Diagnosis of bronchiolitis was originally based on criteria published by Court,15 but was also consistent with other later published criteria.16 Demographic details at age 18 are given in table 2.

Structured questionnaires were completed at all follow-up time points to record demographic, hereditary and clinical symptoms of allergy or wheeze. A clinical examination was also performed. The investigations performed at each assessment are summarised in table 1. Details on skin prick tests (SPT) and serum IgE tests are given in our previous publications.11–13 The specific IgE assay method used (ImmunoCAP) was consistent throughout the study period.

Spirometry was performed at baseline from age 7 and post-bronchodilator at ages 13 and 18 years, according to American Thoracic Society recommendations,17 and expressed as standard deviation scores (z-scores) using recently published normative data.1819 Inhaled long-acting β2 agonists were withheld for 24 h and short-acting β2 agonists or cromoglycates for 6 h prior to testing. Inhaled corticosteroids (ICS), if used, were not discontinued. Isocapnic dry air hyperventilation challenge (DACh) was performed after baseline spirometry at age 13 and 18 years (described in more detail in the online supplement). At age 18, prior to these lung function tests, the fraction of nitric oxide in expired air (FeNO) was measured in duplicate at an expiratory flow of 50 ml/s using the Niox Mino (Aerocrine, Stockholm, Sweden) and the mean FeNO result reported.

Multiple-breath washout (MBW) was performed before spirometry using sulfur hexafluoride (SF6) as the marker gas and a mass spectrometer for gas analysis and as previously described in detail elsewhere.20 Lung clearance index (LCI) was calculated as the number of lung volume turnovers (ie, the cumulative expired volume divided by the functional residual capacity) needed to lower the end-tidal tracer gas concentration to 1/40th of the starting concentration. A high value of LCI thus indicates abnormal ventilation distribution. The mean LCI result from three MBWs in each subject was reported. In a previous study including healthy subjects, the mean, SD and upper limit of normality (ULN; mean + 1.96 SD) for LCI were 6.33, 0.43 and 7.17, respectively.20

At each follow-up time point, asthma was defined as ≥3 episodes of physician-verified wheeze and RW as ≥3 episodes of parent-reported wheeze. Definitions of the wheezing patterns over time are shown in table 3. Allergic rhinoconjunctivitis (ARC) or clinical allergy was defined as rhinitis/conjunctivitis occurring at least twice following exposure to a particular allergen and unrelated to infection. Atopic dermatitis (AD) was defined as a pruritic, chronic or chronically relapsing non-infectious dermatitis. Allergic sensitisation implied occurrence of IgE antibodies estimated by SPT and/or serum IgE tests. Current disorder denotes symptoms over the last 12 months. Active and passive smoke exposures, the presence of pets in the household, and atopic or allergic heredity were assessed by questionnaire. A positive history of atopic disease (AD, ARC or asthma) in first-degree relatives (parents or siblings) was based on physician diagnosis.

Statistical analyses

Yates' corrected χ2 test was used for estimation of differences in prevalence among groups and subgroups. One-way ANOVA was used for parametric continuous variables and Tukey HSD for subsequent group comparisons, if the overall F-test was significant. A Mann–Whitney test was used to test group differences for non-parametric continuous variables. 95% CIs for mean and median differences were calculated. χ2 tests for trend were used to assess the combined influence of group allocation (RSV vs control) and a parental history of physician-diagnosed asthma on asthma/RW, ARC and sensitisation at age 18. Pearson correlation tests were used to assess correlation of parametric data, while the Spearman rank test was used for non-parametric data. Kaplan–Meier survival analysis was used to compare time free from diagnosis of asthma, ARC, positive Phadiatop test or positive SPT to any perennial allergen at a follow-up station; p values based on Mantel-Cox log ranks were calculated. Multivariate logistic regression analyses were performed to establish risk factors for asthma/RW, asthma alone and ARC at age 18. ORs with 95% CI and p values were reported. Additionally, conditional logistic regression tests were performed to adjust for any degree of residual confounding due to the constitution of the groups during the study period. SPSS Version 15.0 software for Windows (SPSS Inc, Chicago, Illinois, USA), SPSS SamplePower 2.0 software and CI Analysis software version 2.1.2 (Trevor Bryant, University of Southampton) were used for the statistical analyses.

Table 1

Study design and proportion of the 47 children in the RSV cohort and the 93 in the control cohort seen at follow-up during the study period

Table 2

Demographic data and family history of the respiratory syncytial virus (RSV) and control groups at age 18

Table 3

Wheeze patterns defined from the presence/absence (+/−) of asthma/recurrent wheeze at four follow-up stations and the numbers of cases in each wheeze pattern group for each cohort

Results

Forty-six of the 47 subjects with RSV and 92 of the 93 controls were followed up to 18 years of age. The demographic details of the two cohorts at age 18 are summarised in table 2, and from age 1–18 years in table E1 in the online supplement. No significant differences were seen at age 18.

Cross-sectional data at age 18

Current asthma/RW was documented in 18 of 46 (39%) subjects with RSV and 8 of 92 (9%) controls (p=0.001). Current asthma alone was found in 15 of 46 (33%) subjects with RSV and 6 of 92 (7%) controls (p<0.001). ARC was diagnosed in 20 of 46 subjects with RSV (43%) and 16 of 92 (17%) controls (p=0.002). No difference was found in the prevalence of AD (5 of 46 (11%) subjects with RSV vs 8 of 92 (9%) controls).

The prevalence of sensitisation determined by SPT was significantly increased in the RSV cohort compared with controls to any animal dander (cat, dog, horse) (33% vs 11%; p=0.005) or any perennial (animal danders and house dust mites (HDM)) (41% vs 14%; p=0.001) (see table E2 in online supplement). A higher prevalence of positive Phadiatop responses was seen in the RSV cohort (56% vs 28%, p=0.005) (see table E2 in online supplement). For the whole cohort, the most commonly identified specific IgE antibodies performed in those testing positive to Phadiatop screening were to ‘any perennial’ (39 of 126, 31%), and were significantly increased in subjects with RSV compared with controls (51% vs 21%; p=0.001).

Airway function data are summarised in table 4. Reduced spirometric airway function (forced expiratory volume in 1 s (FEV1), ratio of FEV1 to forced vital capacity (FVC) and forced expiratory flow at 25–75% FVC (FEF25–75)) was documented in the RSV cohort compared with controls, but LCI did not differ. AHR, bronchodilator response and blood eosinophil cell counts were greater in the RSV cohort than in the controls, but FeNO was not. FEV1/FVC and FEF25-75 remained lower in the RSV cohort after bronchodilation. Subjects with RSV had lower spirometry results than the controls, irrespective of current asthma/RW diagnosis (table 5). Spirometry findings were similar in controls with or without current asthma/RW. LCI was significantly raised in subjects with RSV (p=0.006) and controls (p=0.033) with current asthma/RW compared with corresponding subjects without asthma (see table E3 in online supplement). Only LCI differed significantly between the controls with and those without current asthma/RW (see table E4 in online supplement). In the RSV cohort, the maximum percentage fall in FEV1 after dry air challenge correlated significantly with LCI (r2 =0.44; p<0.001), FeNO levels (r2=0.26; p<0.001) and with blood eosinophil counts (r2=0.12; p=0.011), but not with spirometry results (see table E5 in online supplement). Among the subjects with current asthma/RW, regression analyses also demonstrated significant relationships between LCI and the maximum percentage fall in FEV1 after dry air challenge (r2=0.26; p=0.005) and between LCI and FeNO (r2=0.29; p=0.003).

Table 4

Airway function and inflammatory markers in RSV and control groups at age 18 years

Table 5

Airway function and inflammatory markers in RSV versus control subjects with or without current asthma/RW at age 18 years

Family history and other risk factors for asthma and ARC at age 18

A family history of asthma and atopy did not differ between the RSV and control cohorts at age 18. A number of risk factors for current asthma/RW, current asthma and current ARC, respectively, were assessed using multivariate logistic regression analyses including both cohorts. For current asthma/RW and for current asthma, the risk factors included were: allocation (RSV/control), gender, domestic furred pets in the first year of life, parental smoking in the first year of life, own smoking at age 18, current ARC of the subjects themselves, physician-diagnosed parental ARC and physician-diagnosed parental asthma.

For current asthma/RW, only RSV (OR 6.2; 95% CI 2.0 to 19.2; p<0.001) and current ARC of the subjects (OR 6.1; 95% CI 2.1 to 18.1; p<0.001) were significant independent risk factors. For current asthma alone, similar results were found (RSV: OR 7.2 (95% CI 2.1 to 23.9; p<0.001); current ARC: OR 4.4 (95% CI 1.4 to 14.0; p<0.001)).

The risk factors included in the analysis for current ARC were the same, except for current ARC of the subjects. Only RSV (OR 3.6; 95% CI 1.6 to 8.5; p=0.003) was a significant independent risk factor. Corresponding conditional logistic regression tests gave similar results (see table E6 in the online supplement).

Numerically, the RSV subgroup with a parental history of asthma had a higher prevalence of asthma/RW, ARC and positive sensitisation than the RSV subgroup without parental asthma, but no significant differences were found (p=0.058–0.308). The statistical power to detect significant differences between the RSV subgroups was <50%. For the two cohorts analysed together, however, the combination of RSV allocation and history of parental asthma resulted in significant trends for disease and sensitisation (figure 1A–D).

Figure 1

Proportion (%) with (A) current asthma/recurrent wheeze, (B) current allergic rhinoconjunctivitis, (C) current positive Phadiatop test and (D) current positive skin prick test to perennial allergens in respiratory syncytial virus (RSV) and control cohorts with respect to heredity for asthma. χ2 test for trend was used in all graphs and error bars denote 95% CI.

Longitudinal data over the entire study period: subjects with RSV versus controls

Kaplan–Meier survival plots for time free from asthma diagnosis, ARC diagnosis, positive Phadiatop test or positive SPT are shown in figure 2A–D. The RSV group had significantly shorter time free from these diagnoses and test findings.

Figure 2

Proportion (%) of subjects in the respiratory syncytial virus and control cohorts who never had (A) an asthma diagnosis at follow-up at ages 3, 7, 13 and 18 years (log rank (Mantel-Cox) χ2 21.0; df 1; p<0.001); (B) an allergic rhinoconjunctivitis diagnosis at follow-up at ages 3, 7, 13 and 18 years (log rank (Mantel-Cox) χ2 15.5; df 1; p<0.001); (C) a positive Phadiatop test at follow-up at ages 3, 7, 13 and 18 years (log rank (Mantel-Cox) χ2 8.3; df 1; p=0.004); and (D) a positive skin prick test at follow-up at ages 3, 7, 13 and 18 years (log rank (Mantel-Cox) χ2 8.0; df 1; p=0.005).

Reduced lung function in the RSV cohort was evident from the age of 7 years (mean (SD) baseline FEF25–75 z-scores in RSV group vs controls at age 7: −0.63 (1.04) vs −0.22 (0.95) (95% CI 0.05 to 0.77; p=0.025); at age 13: −0.58 (0.77) vs −0.12 (0.78) (95% CI 0.17 to 0.74; p=0.002). At ages 13 and 18 a greater fall in FEV1 after DACh was seen in subjects with RSV.13

Wheeze patterns up to age 18 years

The persistent/relapsing asthma/RW pattern occurred more frequently in the RSV cohort with fewer non-wheezers in the RSV group (table 3). No difference in other wheeze patterns was seen. All 14 individuals with a persistent/relapsing asthma/RW pattern in the RSV cohort had current symptoms on at least two time points (seven on all four and six on three occasions). Asthma/RW was noted from age 3 in 11 of 14 subjects in this group. The one individual in the control group with persistent/relapsing asthma had symptoms at all occasions from age 7. Nine subjects with RSV (7 with persistent/relapsing asthma/RW and 2 with late onset asthma/RW) and 3 control subjects with late-onset asthma/RW were treated with ICS (p=0.002). The persistent/relapsing group (14 RSV and 1 control) was characterised by the highest prevalence of co-existing ARC (figure 3A) and sensitisation (figure 3B). The significantly higher prevalence of sensitisation was evident at ages 3, 7 and 13, and similarly for ARC at age 7 and 13 (data not shown).

Figure 3

Proportion (%) of subjects with (A) allergic rhinoconjunctivitis in different wheezing phenotypes (overall χ2 p<0.001, **p<0.01, ***p<0.001 vs the no-wheeze phenotype) and (B) positive Phadiatop test in different wheezing phenotypes (overall χ2 p<0.001, ***p<0.001 vs the no-wheeze phenotype). Error bars denote 95% CI. RSV, respiratory syncytial virus.

At 18 years of age the persistent/relapsing wheeze group had significantly lower airway function (FEV1, FEV1/FVC and FEF25–75 z-scores), higher LCI and AHR than the non-wheeze group (table 6). Bronchodilator response, FeNO and blood eosinophil counts were significantly increased compared with never wheezers only in the persistent/relapsing asthma/RW group.

Table 6

Airway function and inflammatory markers in the different wheeze pattern groups for the entire cohort at age 18 years

Discussion

The previously reported over-representation of asthma (and asthma/RW), clinical allergy and allergic sensitisation in this RSV cohort persisted into early adulthood. The high prevalence of current asthma/RW in the RSV cohort was due to a high proportion of subjects with early-onset allergy-associated wheezing persisting through childhood and adolescence, and was accompanied by reduced airway function, elevated FeNO and eosinophil counts. A history of hospitalisation for RSV bronchiolitis was the only significant risk factor identified at age 18 for current asthma/RW, asthma alone or ARC.

At age 18, spirometry results were reduced in subjects with RSV with or without current asthma/RW compared with corresponding controls, but subjects with RSV without current asthma did not show evidence of small airway dysfunction as measured by LCI. Spirometry results showed no relationship with AHR or markers of allergic inflammation. In contrast, small airway dysfunction (LCI) correlated with current asthma, AHR and FeNO, a marker of ongoing allergic airway inflammation.

This controlled follow-up study and its findings are unique. It describes the development of asthma and allergy prospectively after severe primary RSV bronchiolitis in the first year life, and has involved several follow-up time points from infancy to young adulthood, all with very high attendance rates. The subsequent high frequency of early-onset allergic asthma persisting to age 18 has not previously been reported. Only two other prospective studies of RSV have documented a high prevalence of early sensitisation comparable to ours at 1 year21 and 3 years of age.22

The debate between a causal or genetic predisposition relationship between RSV and asthma has continued for decades and cannot be answered by this study. Two recent large registry-based studies have attempted to answer this question, with each declaring opposite conclusions.2324 The weaknesses in the study designs and the difficulties in answering this question have been highlighted in a recent editorial.25 RSV-positive infants with wheezing may include both those with primary RSV LRTI or those with an existing non-atopic or atopic wheezing disorder exacerbated by RSV. In our cohort, 91% of the index subjects were 6 months or younger when admitted and only one had a previous history of lower airway symptoms, which strongly suggests that our cohort comprises subjects with early primary severe RSV LRTI. It is thought that early infancy, with a relatively Th-2 skewed immune system, may constitute a particularly vulnerable period of life for subsequent development of asthma following severe viral LRTI.3 Viral airway infections and atopy may interact in a multiplicative way to promote asthma development in young childhood.26 In addition, predisposition to both early severe RSV bronchiolitis and allergic sensitisation may be related to the interleukin (IL)-13/IL-4 gene locus.27 Potential hereditary susceptibility to RSV bronchiolitis has also been reported.28

Potential confounding factors do exist within our cohort. Mallia and Johnston29 have previously suggested that the higher rate of asthma in our RSV cohort compared with controls could be due to selection bias generated by the selection of controls contemporaneously from the same child healthcare centres as the index cases. Theoretically, our selection process during an ongoing RSV epidemic could generate controls with a lower susceptibility for severe RSV bronchiolitis and a lower risk for subsequent allergic asthma if severe RSV bronchiolitis and subsequent allergic asthma are markers of the same genotype. There are two main reasons why we do not feel that selection bias significantly affected our results. First, the RSV cohort had the same ARC and asthma parental/sibling heredity prevalence as the controls, and the RSV and control cohorts had a similar prevalence of AD. Second, the control group had a similar prevalence of current asthma/RW, ARC and positive SPT at age 13 to a recent population-based sample of 12–13-year-old Swedish children living in an overlapping geographical area (8.5%, 17% and 22.5%, respectively).3031 Our study was intended to be a descriptive study and, as such, was not powered to detect differences in hereditary rates of allergic disease/asthma between asthmatic and non-asthmatic index cases. Fifty index cases in each group would have been required to achieve adequate power. Nevertheless, there are indications that early severe RSV bronchiolitis is associated with a greater increased risk for subsequent asthma, clinical allergy and allergic sensitisation when combined with a history of parental asthma (figure 1 A–D).

The FEV1/FVC ratio and the FEF25–75, the most reliable and the most sensitive spirometric indices of airway obstruction, respectively, were reduced both at baseline and post-bronchodilator in our RSV cohort. Airway obstruction was most pronounced in subjects with a persistent/relapsing wheezing pattern at age 18, and this pattern could be traced back to ages 7 and 13 years. Interestingly, spirometric indices were also reduced in subjects with RSV without current asthma. These changes may reflect premorbid airway function32 or the presence of airway remodelling due to a severe RSV infection occurring during a critical period of lung development, with or without subsequent interaction with early allergic sensitisation.233 Young age at first wheezing episode is an important risk factor for subsequent persistent asthma in sensitised cohorts.34

In the present study LCI was increased in both RSV and control subjects with current asthma, while spirometry results were impaired only in the subjects with RSV. Increased LCI can result from non-uniformity of ventilation distribution among lung regions sharing branch points in the conducting airway zone or even more peripherally in the vicinity of the terminal bronchioles.3536 These results suggest that residual small airway dysfunction is not a result of RSV infection per se but is related to a subsequent diagnosis of asthma. Peripheral airway obstruction and the patchy distribution of airway disease are both recognised characteristics of asthma.937 Our findings are consistent with other studies documenting raised LCI, despite normal spirometry, in subjects with current mild asthma aged 5–15 years,38 and normal small airway function in an RSV cohort without asthma followed up at age 10 using single-breath nitrogen washout.5 The marked alveolarisation and completion of distal airway formation occurring during the first three years of life may compensate for the more peripheral airway sequelae of early severe RSV infection, while more proximal changes persist. Increased AHR was present in the subjects with asthma, although not to the magnitude previously reported in other studies, in part due to the decision not to withhold ICS prior to testing. A fall in FEV1 of 10% or more (mean 22.1%) was seen in 8 of 41 subjects with RSV and in 5 of 86 tested controls (mean 20.1%; p=0.038; χ2 with Yates' correction). Despite this, a significant correlation of LCI but not spirometry with AHR was ween in our RSV cohort and confirms the important findings of Downie et al.8 LCI but not spirometry was also significantly correlated with ongoing airway inflammation (FeNO).

In summary, this study shows that severe primary RSV bronchiolitis in the first year of life is frequently followed by allergic asthma persisting into early adulthood. Subjects with RSV with and without current asthma/RW have reduced airway function as measured by spirometry. Ventilation inhomogeneity, a measure of small airway function, is normal in subjects with RSV without current asthma but is linked to current asthma, AHR and ongoing airway inflammation. Our findings suggest that early severe RSV bronchiolitis has lifelong consequences of allergic asthma and airway remodelling.

Acknowledgments

The authors thank Mrs Gunilla Holmgren Wallmyr for skilful assistance at all follow-up visits in this study and Mr Salmir Nasic for statistical advice.

References

Abstract

To determine the relation between obesity and new-onset asthma among school-age children, the authors examined longitudinal data from 3,792 participants in the Children’s Health Study (Southern California) who were asthma-free at enrollment. New cases of physician-diagnosed asthma, height, weight, lung function, and risk factors for asthma were assessed annually at five school visits between 1993 and 1998. Incidence rates were calculated, and proportional hazards regression models were fitted to estimate the adjusted relative risks of new-onset asthma associated with percentile of body mass index (weight (kg)/height (m)2) and indicators of overweight (>85th body mass index percentile) and obesity (>95th body mass index percentile). The risk of new-onset asthma was higher among children who were overweight (relative risk (RR) = 1.52, 95% confidence interval (CI): 1.14, 2.03) or obese (RR = 1.60, 95% CI: 1.08, 2.36). Boys had an increased risk associated with being overweight (RR = 2.06, 95% 1.33, 3.18) in comparison with girls (RR = 1.25, 95% CI: 0.83, 1.88). The effect of being overweight was greater in nonallergic children (RR = 1.77, 95% CI: 1.26, 2.49) than in allergic children (RR = 1.16, 95% CI: 0.63, 2.15). The authors conclude that being overweight is associated with an increased risk of new-onset asthma in boys and in nonallergic children.

allergy and immunology; asthma; body mass index; body weight; child; obesity

Abbreviations: BMI, body mass index; CI, confidence interval; ETS, environmental tobacco smoke; RR, relative risk.

Received for publication November 13, 2002; accepted for publication March 12, 2003.

Asthma is a large and growing threat to children’s health and well-being (1). In some communities, the prevalence of asthma among school-age children exceeds 25 percent, and prevalence has been rapidly rising in many regions of the developed world (2–6). Although asthma is the subject of intense research efforts, the etiology of asthma and the reason for the increase in prevalence have yet to be firmly established (1, 7–10).

Much of the focus of childhood asthma research has been on atopy and the development of allergic responses to common indoor allergens; however, atopic pathways do not appear to contribute to a substantial portion of cases, and other etiologic pathways that involve nonallergic mechanisms are likely to be involved (7, 8, 11–14). Exposures of interest whose effects may be partly mediated by nonallergic pathways include in-utero and postnatal tobacco smoke exposure, ozone, infant feeding practices, and viral infections (9, 12, 13, 15–21). An emerging body of evidence suggests that obesity may play a role in the development of childhood asthma through nonallergic pathways (1, 14, 22–28).

It has long been recognized that obesity is more common among children with asthma, and associations between asthma and high body mass index (BMI) (weight (kg)/height (m)2) have been observed in cross-sectional studies of adults and children (1, 14, 22, 29). These associations have been explained as evidence that asthma causes obesity due to a lack of physical activity among children with asthma; however, this interpretation has been challenged by the results of recent longitudinal studies. In adults, obesity is associated with an increased risk of asthma in prospective studies, especially among women (23, 26, 27). In girls, becoming overweight or obese between the ages of 6 and 11 years has been found to increase the risk of developing new asthma and to increase bronchial responsiveness during adolescence (25). Additional longitudinal studies of children are needed to define the temporality of the association between asthma and obesity and to determine whether the association with obesity is restricted to girls or to nonallergic children (30).

The Children’s Health Study, a longitudinal study of respiratory health among school-age children in 12 Southern California communities, provided us with an opportunity to investigate whether being overweight or obese is an antecedent condition associated with increased risk of newly diagnosed asthma and whether any risk associated with being overweight or obese varies by sex or allergy status (3). We examined the association of new cases of physician-diagnosed asthma and the development of obesity using data collected at yearly assessments between 1993 and 1998 from a cohort of 3,792 children who were asthma-free at study enrollment.

MATERIALS AND METHODS

The Children’s Health Study is a prospective study of the determinants of children’s respiratory health (3). Children were recruited in 1993 from fourth-, seventh-, and 10th-grade classes in public schools in 12 Southern California communities. A second cohort of fourth graders was recruited in 1996. At study entry, the parents or guardians of each participating student provided written informed consent and completed written questionnaires that provided information on sociodemographic factors, history of respiratory and allergic illnesses and their associated risk factors, exposures, including smoking by household members, and household characteristics. Children with any lifetime history of asthma at study entry were considered to not be at risk and were excluded from the analysis. This resulted in a cohort of 3,792 children (1,993 girls and 1,799 boys). Wheezing was defined as any lifetime history of wheezing at study entry. Children with a history of wheezing but no diagnosis of asthma were considered to be at risk for a new diagnosis of asthma and were included in the study.

Children were assessed annually during school visits until high school graduation. The analyses described here included data collected at five visits between 1993 and 1998 for fourth and seventh graders who entered the study in 1993 and data collected at three visits for 10th graders between 1993 and 1995 and the second cohort of fourth graders between 1996 and 1998. During each annual assessment visit to schools, children completed an update questionnaire and interview that included items on physician diagnoses of asthma, other respiratory symptoms, and recent exposure history. An incident asthma case was defined as a new physician diagnosis of asthma during the time between follow-up assessments. Date of diagnosis for incident asthma cases was assigned as the midpoint of the period between follow-up assessments.

Height, weight, and lung function were measured annually using standard protocols. Because the relation between BMI and obesity changes with age and varies by sex, we categorized BMI into age- and sex-specific percentiles based on the Centers for Disease Control and Prevention BMI growth charts using 1-month age intervals (31). Overweight was defined as BMI greater than the age- and sex-specific 85th percentile and obesity as BMI greater than the 95th percentile. Lung function was measured as previously described (32). Age- and sex-specific percent predicted values for forced vital capacity, forced expiratory volume in 1 second, and forced expiratory flow rate between 25 percent and 75 percent of forced vital capacity were calculated on the basis of lung function-height relations in the study population.

Potential cofounders or effect modifiers were identified by review of the literature and preliminary analyses and included age, sex, race/ethnicity, health insurance, community of residence, parental history of asthma and allergies, birth weight, humidifier use, history of wheezing, history of allergy, participation in team sports, personal smoking, household environmental tobacco smoke (ETS) exposure, household pets and pests, puberty, and lung function level. Wheezing with exercise was defined as a parental report of wheezing with exercise, and personal and parental family histories of allergy were based on self-reported physician diagnoses of allergies on the baseline questionnaire, including eczema, hay fever, and symptoms of allergy to pets and dust. Allergic rhinitis was defined by parental report of hay fever or nasal allergy. Age of puberty was defined as the age of peak height velocity and was assigned as the midpoint of the interval between annual examinations where maximum growth occurred. Recent inhaled medication use and personal smoking habits were ascertained annually by private interview by field team members.

Statistical methods

Incidence rates for new cases of physician-diagnosed asthma were calculated using the number of new cases divided by the person-years at risk over a 4-year period of follow-up. Stratified rates were calculated for boys and girls, as well as age- and sex-specific BMI percentile categories at study entry. The numbers of new cases and person-years at risk within each stratum were determined and summed over the follow-up period. For cases, the number of person-years at risk was fixed at 0.5 for the year in which asthma was diagnosed.

To further investigate the relations between BMI and new physician diagnoses of asthma with adjustment for the effects of confounding variables, we fitted Cox proportional hazards regression models. We used stratified baseline hazards allowing for sex and age strata in the analysis of the entire cohort and in analyses stratified by allergy status, and we used baseline hazards stratified on age in the analyses of subgroups defined by sex. We investigated associations of new-onset asthma with obesity by including BMI percentiles and overweight and obesity status in the models as time-dependent variables using a 1-year lag period. For sensitivity analyses, we used 2-year lag periods for BMI, overweight, and obesity, as well as fixed BMI and obesity categories at study entry. Additional time-varying covariates included puberty status and ETS exposure variables. In the final regression models, the association between asthma and BMI was adjusted for age, community, race/ethnicity, wheezing, and sex-specific effects of allergy. We found little evidence for confounding (<10 percent change in BMI estimates) by other covariates, including lung function level (forced expiratory volume in 1 second percent predicted). We assessed the heterogeneity of associations among subgroups by fitting separate models in each subgroup and by including interaction terms in regression models applied to the entire cohort; the statistical significance of differences was tested by likelihood ratio test. We conducted sensitivity analyses by limiting the case definition to children who reported inhaled medication use and by restricting the cohort to children without a history of wheezing at study entry. All analyses were carried out using SAS software (SAS Institute, Inc., Cary, North Carolina). Unless otherwise noted, all hypothesis testing was conducted assuming a 0.05 significance level and a two-sided alternative hypothesis.

RESULTS

At study entry, participants ranged in age from 7 years to 18 years (table 1). The majority of children were non-Hispanic White or Hispanic White. Approximately 20 percent of the children had a history of physician-diagnosed allergic rhinitis. A lifetime history of any wheezing was reported for 24 percent of boys and 21 percent of girls. Few children smoked, but approximately 18 percent had a lifetime history of ETS exposure. More boys than girls were overweight or obese at study entry.

Risk factors for a new physician diagnosis of asthma varied by sex (table 2). Among girls, a history of allergy in the mother and a history of wheezing or exercise-induced wheezing were risk factors for new-onset asthma. Among boys, risk factors for new-onset asthma included other race/ethnicity, puberty, family history of asthma, history of allergy or allergic rhinitis, history of wheezing or wheezing with exercise, household ETS exposure, and household humidifier use. The effects of puberty, wheezing status, family history of asthma, allergy and allergic rhinitis history, and ETS exposure differed significantly in boys compared with girls. Health insurance, history of allergy in the father, household pets and pests, participation in team sports, parental income, and education were not significant risk factors in this analysis.

Over the 4-year period of follow-up, 288 new cases of asthma were diagnosed. The overall crude incidence rate was 24.6 per 1,000 person-years at risk. Girls had a higher incidence rate than boys (table 3). Crude rates of new-onset physician-diagnosed asthma were higher in overweight and obese children than in children of normal weight. Although the elevated rates were apparent in both overweight and obese boys and girls, the patterns of rates varied by sex. Incidence rates of asthma were generally higher for normal-weight girls than for normal-weight boys. In contrast, obese boys had higher rates of asthma than obese girls. As a result of this pattern of rates, boys had a larger difference between the obese and normal-weight groups than did girls. Among girls, the rates in the overweight group at study entry were higher than the rates in the obese group. Rates in the lowest percentiles of BMI were not substantially different from those among normal-weight children.

Results from the multivariable models showed that the relative risk for new physician-diagnosed asthma was increased in the upper BMI percentiles (table 4). Overweight and obesity were both associated with increased risk of a new diagnosis of asthma (relative risk (RR) = 1.52 (95 percent confidence interval (CI): 1.14, 2.03) and RR = 1.60 (95 percent CI: 1.08, 2.36), respectively).

As is suggested by the patterns of crude incidence rates for BMI percentiles, we saw some evidence that the adjusted relative risks comparing overweight or obese children with normal-BMI children were larger in boys than in girls (p = 0.09). Obese boys had an adjusted relative risk of 2.29 (95 percent CI: 1.35, 3.88) compared with nonobese boys. Among girls, the relative risk associated with obesity (RR = 1.10, 95 percent CI: 0.60, 2.05) was smaller than that among boys, and none of the relative risk estimates for BMI percentile were statistically significant. To assess the potential for misclassification of new-onset asthma, we conducted a sensitivity analysis that used a case definition that required a report of inhaled medication use in addition to a new diagnosis of asthma. When we restricted cases to children who had used inhalers recently, the risk estimates changed little. For example, in the analysis restricted to those who had recently used inhalers, the relative risk for obese children compared with nonobese children was 2.31 (95 percent CI: 1.22, 4.37) for boys and 1.10 (95 percent CI: 0.49, 2.47) for girls. The associations with being overweight followed the same patterns and were of approximately equal magnitude as those for obesity. In further sensitivity analyses that restricted the cohort to children without a history of wheezing, the risk estimates did not change substantially in comparison with the estimates from the analyses that included all children.

The associations of overweight with new-onset asthma were significant and were larger in nonallergic children (RR = 1.77, 95 percent CI: 1.26, 2.49) than in allergic children (RR = 1.16, 95 percent CI: 0.63, 2.15) (p = 0.03) (table 5). There was little evidence for an effect of overweight in allergic children. The asthma risk associated with being overweight was larger and statistically significant in nonallergic boys.

We found that the effects of being overweight were larger among boys than among girls in models that assessed interactive effects of obesity and sex in children with different allergy statuses (p < 0.03) (data not shown). The relative risk for overweight in boys was 1.93 (95 percent CI: 1.05, 3.59) times higher than that in girls (p = 0.03). In addition, the effect of being overweight was approximately half (RR = 0.47, 95 percent CI: 0.23, 0.94) that in allergic children compared with nonallergic children (p = 0.01). We found little evidence for differences in the relative risks for being overweight or obese by age, puberty status, early onset of puberty, family history of asthma or allergy, wheezing status, race/ethnicity, participation in team sports, and household ETS exposure.

In sensitivity analyses, adjustment for health insurance, parent/guardian educational attainment, parental income, parental history of asthma and allergies, birth weight, humidifier use, history of exercise-induced wheezing, history of allergy, participation in team sports, personal smoking, household ETS exposure, household pets and pests, puberty status, and lung function level did not substantially change any of the risk estimates for BMI percentiles in either boys or girls. We attempted to assess the effects of change in obesity over the follow-up period by modeling the annual change in obesity status; however, too few nonobese children became obese in the subsequent year for an informative analysis. The associations with new-onset asthma were slightly reduced in magnitude when fixed BMI and obesity categories at study entry or 2-year lagged BMI categories were used in the models instead of the 1-year lagged time-varying BMI percentiles.

DISCUSSION

Understanding the relation between obesity and asthma in children may be an important step in clarifying the etiology of childhood asthma. Our findings support the hypothesis that being overweight or obese is associated with increased risk of new-onset asthma in children. A growing body of evidence supports the possibility that obesity increases the risk of new-onset asthma (1, 22, 23, 25–28). Because the vast majority of studies on the relation between asthma and obesity have been cross-sectional, the temporal relation between obesity and asthma onset is not clear. Three recent longitudinal studies in adults have suggested that obesity precedes asthma and is associated with an increased risk in adults that has been more consistently observed in women and appears to be larger in women than in men (23, 26, 27). In the Nurses’ Health Study cohort, obese women and women who gained weight after 18 years ofage were at significantly increased risk of developing asthmaduring the 4-year follow-up period (23). Among participants in the two Canadian National Population Health Surveys, asthma incidence was associated with BMI in females but not in males (33). In the Coronary Artery Risk Development in Young Adults Study, gain in BMI predisposed women to new asthma diagnosis, and decreased physical activity did not explain the association of weight gain with asthma (26). To date, one prospective study has examined the association between obesity and new-onset asthma in children (25). In girls, becoming overweight or obese between 6 and 11 years of age increased the risk of developing new asthma and increased bronchial responsiveness during adolescence (25). Obesity was not associated with asthma incidence in boys.

Our findings are consistent with an increased risk of asthma among overweight and obese children; however, we found evidence that the risk was largely restricted to boys. We lacked data to explore the inconsistency in results for boys and girls between studies; however, chance, differences in study populations, the age distribution of participants, or unrecognized confounding or effect modification may have contributed. After we accounted for sex differences in the effects of allergy and differences in the effects of BMI percentile by allergy status, the relative risks for overweight and obesity were larger in boys than in girls (p = 0.03). Differences in allergy status and in the age distribution of our population, which was slightly older than the cohort examined by Castro-Rodriguez et al. (25), might explain the discrepant results, particularly if the sex differences in the relation of obesity with asthma reverse in late adolescence. Because obese women appear to be at higher risk than men, a reversal of the sex difference in late adolescence is not a likely explanation. In addition, we found no variation in the effect of obesity over the age range in our study. Risk factors for asthma varied by sex in our cohort, suggesting that different patterns of confounding or effect modification may have occurred, especially in relation to allergy status. We assessed a wide spectrum of potential confounders in both boys and girls, including lung function level, physical activity, ETS exposure, and puberty status, and found that adjustments did not substantially change the sex-specific relative risks. Unrecognized bias or differences in exposures or doses that interact with obesity to affect asthma risk could also play a role in explaining the inconsistent results in boys and girls.

A better understanding of the mechanisms for the effects of BMI on asthma risk may contribute to interpretation of the results from epidemiologic studies. It has been suggested that the association between obesity and asthma results from mutual correlation of obesity and asthma with common etiologic factors (10, 34, 35). Change in lifestyle may explain the association, as well as the co-occurrence of increasing prevalence for these conditions. Obesity is associated with a lack of physical exercise and a diet high in calories, and activity levels and dietary habits may be related to the onset of childhood asthma (10, 24, 34, 35). Thus, some aspect of the lifestyle associated with obesity, such as more time spent indoors, may be the etiologically important factor for new-onset asthma in some communities.

Obesity is associated with a large number of changes in physiology that may mediate the relation of obesity with asthma (24). Obese persons show systemic inflammation that appears to play a role in the etiology of nonatopic conditions, including cardiovascular disease, diabetes, and potentially asthma. Adipose tissue is a source of proinflammatory cytokines and chemokines such as interleukin-6, leptin, interleukin-18, and tumor necrosis factor-α. An increase in circulating levels or local concentrations of proinflammatory cytokines has the potential to enhance pulmonary inflammation, which is a key component of asthma pathophysiology. Because obesity is not clearly associated with allergy, obesity may enhance noneosinophilic inflammatory pathways that increase the risk of nonatopic asthma. The effects of obesity may also be mediated by changes in airway function, since obesity and weight change have been prospectively associated with increased bronchial hyperresponsiveness in asthmatic children as well as in nonasthmatic children. The combined effect of increased bronchial hyperresponsiveness and the proinflammatory milieu in obese subjects may set the stage for the onset of asthma (36–40).

In the present study, the incidence rate of physician-diagnosed asthma was higher than incidence rates of childhood asthma reported in earlier decades (41–43). Incidence rates for young adults during recent periods have been reported to be in the range of 3–7 cases per 1,000 person-years, with higher rates in women than in men (26, 44, 45). While incidence rates for children have been reported to be lower in earlier birth cohorts than in Children’s Health Study children (41–43), incidence rates in more recent cohorts are comparable to those in the present study. In a cohort of children in Tucson, Arizona, the cumulative incidence of newly diagnosed asthma was 12.0 percent, which is approximately equal to the cumulative incidence in the present study (46). In the British 1958 birth cohort, the average annual incidence of new cases was 26 per 1,000 person-years, 11 cases per 1,000 person-years, 7.1 cases per 1,000 person-years, and 7.6 cases per 1,000 person-years, respectively, over the four age periods examined (0–7 years, 8–11 years, 12–16 years, and 17–23 years) (47, 48). Among 7- and 8-year-old children residing in northern Sweden during 1996 and 1997, the incidence of physician-diagnosed asthma was nine cases per 1,000 person-years (49). Among eighth grade students residing in northern Sweden, the asthma incidence rate was 11 per 1,000 person-years in 1991 (50). In a third study conducted in Sweden between 1990 and 1993, the yearly incidence of asthma in 16- to 19-year-olds was 13 per 1,000 person-years (51). The yearly incidence of asthma in 15 European countries increased more than twofold between the 1946–1950 birth cohort and the 1966–1971 birth cohort (52).

Our cohort entry criteria and ascertainment of new cases of asthma were based on self-reports of physician diagnosis, a process that might have affected incidence rates and relative risk estimates. Physician assessment of asthma has been recommended and has been widely used as a method of classifying asthma status in epidemiologic studies (4, 53). Differences in access to care and differences in practice among physicians have the potential to influence asthma diagnoses (54). Misclassification of asthma status at study entry did not appear to bias our results. We based this conclusion on the fact that exclusion of cases diagnosed in the first year, a period when unrecognized prevalent cases are likely to be diagnosed, did not substantially change the point estimates for obesity and overweight. We found that adjustment for factors that mediate access to care, including parental income, education, and medical insurance, did not explain our results. This indicates that differential access to care in obese and nonobese children did not substantially bias our results. To investigate the role of a past history of wheezing on physician diagnosis of asthma, we examined the risk of a new diagnosis in children with and without a history of wheezing and found that BMI associations showed little variation between groups. This indicates that any bias from initial symptoms was likely to have been small.

On the basis of the assumption that variation in practice patterns is likely to be larger between communities than within a community, we indirectly assessed the role of physician variation between communities by accounting for differences in community of residence; we found little change in the associations with BMI. To further investigate the potential for bias from variations in medical practice, we conducted analyses restricting cases to children who had recently used inhaled medication; we found little change in the risk associated with obesity. Because the associations with BMI were apparent in the group of cases for which the diagnosis was most certain, our results are unlikely to be explained by variation in diagnosis. Furthermore, underdiagnosis of asthma in overweight and obese children may occur, since physicians may inaccurately attribute a child’s postexercise wheezing or shortness of breath to a general lack of physical fitness related to the child’s weight. However, this practice would result in a bias toward the null and therefore would not explain our findings. Because we did not follow the cohort from birth, we could not determine whether a new diagnosis represented an incident occurrence of asthma or a second occurrence of asthma that had first occurred during infancy. We also used parental reports of physician-diagnosed allergy to classify children’s atopic status. Because this approach is likely to underestimate the occurrence of atopy compared with skin testing, the associations of overweight with asthma are likely to have been overestimated in the allergic group of children, lessening the difference in risk between allergic and nonallergic children.

Our findings may have public health significance, since the increasing prevalence of overweight and obesity among children may be an important contributor to the increasing incidence and prevalence of asthma. The prevalence of overweight and obesity among children has been rapidly increasing over the past 20 years—the same period in which the epidemic increase in asthma prevalence has occurred. In the 10 years between the Second and Third National Health and Nutrition Examination Surveys, the prevalence of overweight in the United States increased by 40 percent (55). If obesity contributes to the incidence of asthma, then the rising prevalence of childhood obesity may contribute to the ongoing asthma epidemic, and we may need to target obesity prevention in our efforts to control the epidemic. Further longitudinal epidemiologic and mechanistic studies are needed to identify the causes of the childhood asthma epidemic.

ACKNOWLEDGMENTS

This study was supported by the California Air Resources Board (contract 94-331), the National Institute of Environmental Health Sciences (grants 5P01 ES09581 and 5P30 ES07048), the Environmental Protection Agency (grant R826708 01-3), the National Heart, Lung, and Blood Institute (grant 5R01 HL6176-04), and the Hastings Foundation.

The authors thank Dorothy Starnes for providing technical support in the preparation of the manuscript.

The statements and conclusions in this report are those of the investigators and not necessarily those of the California Air Resources Board. The mention of commercial products, their sources, or their use in connection with material reported herein is not to be construed as either an actual or an implied endorsement of such products.

TABLE 1.

Selected characteristics of participants with no history of physician-diagnosed asthma at study entry, Children’s Health Study, 1993–1998

Characteristic Girls Boys 
(n = 1,993) (n = 1,799) 
No. No. 
Race/ethnicity 
White 1,132 56.8 1,070 59.5 
Hispanic 574 28.8 503 28.0 
Asian 108 5.4 105 5.8 
Black 112 5.6 69 3.8 
Other 67 3.4 52 2.9 
Age (years) at study entry 
7–9 1,045 52.4 894 49.7 
10–11 292 14.7 367 20.4 
12–14  365 18.3 277 15.4 
15–18 291 14.6 261 14.5 
Postpubertal at study entry 
Yes 298 15.2 248 14.0 
No 1,663 84.8 1,519 86.0 
Health insurance 
Yes 1,595 82.5 1,482 85.3 
No 338 17.5 256 14.7 
Family history of asthma 
Yes 278 15.3 247 14.9 
No 1,544 84.7 1,414 85.1 
History of allergy in mother 
Yes 537 28.6 496 29.0 
No 1,338 71.4 1,215 71.0 
History of allergy in father 
Yes 429 23.8 383 23.5 
No 1,371 76.2 1,246 76.5 
History of allergy 
Yes 418 22.1 403 23.6 
No 1,470 77.9 1,306 76.4 
History of wheezing 
Yes 404 21.5 404 24.0 
No 1,471 78.5 1,281 76.0 
History of allergic rhinitis 
Yes 431 22.4 345 19.8 
No 1491 77.6 1,397 80.2 
Wheezing with exercise 
Yes 70 3.8 64 3.8 
No 1,793 96.2 1,618 96.2 
Personal smoking 
Yes 59 3.0 60 3.3 
No 1,934 97.0 1,739 96.7 
Household ETS* exposure 
Yes 351 18.1 308 17.7 
No 1,587 81.9 1,436 82.3 
Humidifier use 
Yes 486 26.0 462 27.0 
No 1,384 74.0 1,251 73.0 
Yearly parental income 
Low (<$15,000) 241 15.8 241 15.8 
Middle ($15,000–<50,000) 673 41.9 652 42.6 
High (≥$50,000/year) 660 41.1 637 

Reprint requests to Dr. Frank Gilliland, Department of Preventive Medicine, Keck School of Medicine, 1540 Alcazar Street, CHP 236, Los Angeles, CA 90033 (e-mail: gillilan@usc.edu).

0 thoughts on “Case Study 50 Allergic Asthma Frank Morgan

Leave a Reply

Your email address will not be published. Required fields are marked *